OCC-vs-ROPP discrepancy in ropp_pp_bending_angle_go

[Ian Culverwell]

User Kjartan Kinch (DMI) has identified another difference between these codes:

In the routine ropp_pp_bending_angle_go there is a call to the routine ropp_pp_satellite_velocities (under point 5.0) and a little later two calls to ropp_pp_geometric_optics (under 6.2 and 6.3) which give as output the geometric-optics bending angles for the L1 and L2 channels. This is the core of the geometric-optics processing and it is called several times in the processing chain. The routine ropp_pp_satellite_velocities takes as input the gns and leo satellite coordinates (in variables r_gns and r_leo). The routine fits the whole time-series with a 5th degree polynomial and differentiates to find the velocities. The routine outputs both fitted velocities (vleo and vgns) and fitted positions (xleo and xgns). Later, ropp_pp_geometric_optics takes as input positions, velocities, and a doppler shift - and outputs the bending angles. The difference is that OCC inputs the fitted positions, while ROPP_PP inputs the original positions. If you replace r_leo and r_gns by xleo and xgns in the calls to ropp_pp_geometric_optics you will get the equivalent of what OCC does. I think I would consider this a minor bug in ROPP_PP unless there's a good reason for this choice. In any case it is a cause of minor discrepancies.

This correction has been made for ROPP5